1The most fundamental questions regarding the history of water mills are when and where was the water mill invented and the relationship historically between the horizontal water mill, that with a horizontal water wheel and the vertical water mill that with a vertical water wheel as described by Vitruvius (De Architectura X, 5, 2). In both cases they are operated by a spindle that goes through a hole in the lower mill stone and turns the upper stone. In the horizontal mill this spindle is connected directly to the water wheel while in the vertical mill because of the change in direction gearing is necessary and a toothed wheel at the end of the axle of the water wheel turns a toothed wheel at the bottom of the spindle that turns the upper millstone. I wish to present here a short survey of what we know about water mills in Israel and while putting this evidence within a wider context I will attempt to show what the evidence from Israel can contribute to answering these questions.

2The pre-industrial water mills of Israel are all horizontal mills (Avitsur 1960 a, b, c). These mills usually had two different devices to set the mill in motion and to stop it. One a board that diverts the water from the water wheel and the other that consists of a rod that raises and lowers a pivoted board that is placed below the water wheel. When the rod raises this board it also raises both the spindle that turns the upper millstone and the upper millstone itself thus controlling the milling process. If the rod is raised far enough the grinding ceases. The rod is moved upwards by inserting a peg into a rectangular hole at its upper end (see Wilson 1995, p. 501 for a description of this device in an ancient water mill at Chemtou in North Africa). In most cases the water is diverted from the stream to a channel which follows the contour and the mill is placed at a point where the channel reaches a point high enough above the stream to provide the drop needed to operate a mill. The water flows down through a wide vertical tube, 4-12 m high, that is placed in a tower and is known as a “Tower Penstock” or an “Aruba Penstock” (in English “penstock is “a channel trough or tube for conveying water from a lake or dam to a water wheel” and Aruba is Hebrew for chimney). Very often there are two or even three mills one above the other the same water operating them all. In Israel and surrounding countries there is little water and few rivers and this method generates the maximum power from small quantities of water. However even where the mills are on rivers they usually also have horizontal wheels. In these cases the water is supplied from behind dams that sometimes create small artificial lakes.

3Before discussing the archaeological evidence from Israel and some ancient written sources only recently brought to the notice of the academic world by Lewis (1997), I will present a short survey of what is known about water mills in other countries and from other written sources.

4Early vertical water mills have been excavated from Britain, France, Greece, and Germany some dating back to the first centuries CE (for general surveys see Wikander 1985, Brun 1998) and nearly all the pre-industrial water mills in western Europe are also vertical (Reynolds 1983). As regards horizontal mills an example from the first century BCE was reported from Bølle in Denmark (Steensberg 1978) and several were published from North Africa including one from the 3^rd-4^th centuries CE from Chemtou (Wilson 1995). The region of Israel and the surrounding countries is also not the only one where all the pre-industrial mills are horizontal (for general surveys see Wilson 1960; Kreiner 2001). This is true for Scandinavia and as a result the horizontal mill is often called the “Norse Mill”. In England and Ireland where the pre-industrial water mills are all vertical, as were those discovered in England from the Roman period, there is evidence that in the middle ages there were horizontal water mills there also (England-Rahtz 1992; Ireland-Lucas 1953; Baillie 1975). Regarding horizontal water mills with Aruba penstocks as well as Israel, Jordan and Lebanon these have also been recorded in Oman (Wilkinson 1980), Iran/Persia (Wulff 1966, p. 280-283; Harverson 1993), Cyprus (Given 2000, p. 225-227) Crete (Rahtz 1981) and Portugal (Dias 1964, fig. 12).

5Until recently the oldest written references that scholars considered relevant to the history of water mills were in Greek, that of Strabo (Geographica XII, 556) who mentions one built by Mithrades king of Pontus that is therefore to be dated to before 65 BCE and the famous epigram telling the maidens that they no longer needed to grind the corn as the water mill would do the work for them (for the Greek with an English translation see Moritz 1958, p. 131). This epigram is usually accredited to Antipater of Thessalonica (1^st c. BCE or 1^st c. CE).

6There is also a Rabbinic source referring to water mills from approximately the same period A ruling appearing in the Tosephta (Tos. Shabbat 1, 28), stipulates that the two famous Rabbinic schools of Hillel and Shammai both forbade the use of water mills on the Sabbath although the two schools agreed to permit the setting up of wine and oil presses before the Sabbath in order to allow them to work on the Sabbath (Mishna Shabbat 1, 9). In the Jerusalem Talmud (JT Shabbat ch. 1, 5, 4/1) there is a detailed discussion as to why the use of water mills was forbidden. One of the reasons given is that the miller “might forget and drive in the peg”, almost certainly the peg, mentioned above that controlled the milling process (for full text and discussion see Frankel 2003b). Hillel and Shammai were active at the end of the first century BCE and the beginning of the first century CE but the Rabbis themselves are not specifically mentioned. Their Rabbinic schools however were not active after the destruction of the second temple in 70 AD so that this injunction must be dated at latest to the first century CE and perhaps earlier. As for the Latin sources, in addition to Vitruvius, Pliny the elder also mentions water mills as being in use “in large parts of Italy” (Naturalis Historiae XVIII, 23, 97). It should be pointed out that in the two Greek sources the water mill is something exceptional suggesting that it was still an innovation, while both the Rabbinic references and that of Pliny suggest that the water mill was already well established. In Pliny and Strabo there is no hint as to what type of mill is referred to but as regards the Greek epigram there are differences of opinion. Early scholars presumed that the mills referred to in the Greek texts were horizontal mills because they are simpler and for this reason the pioneers in the research of ancient mills, Bennet and Elton (1899, 2, 1-10) actually called the horizontal mill “the Greek Mill”. However both Lewis (1985) and Wilson (1995, 505), basing their argument on the Greek text are of the opinion that the Greek epigram refers to a vertical mill. Vitruvius certainly describes a mill with a vertical wheel, and as we have already shown the Talmud almost certainly refers to a horizontal mill.

7It is usually presumed that the simpler mills with horizontal wheels that do not need gearing preceded the more sophisticated Vitruvian mills with vertical wheels and gearing. This plus the facts that all the pre-industrial mills in Israel and its surroundings have horizontal wheels, and that there is an early Rabbinic reference to water mills, led Shmuel Avitzur (1960a) to suggest that the Greek epigram was not written by Antipater of Thessalonica but by Antipater of Sidon (c. 130 BCE) and that the water mill originated in Phoenicia.

8There are however other ancient written sources that refer to water mills and that are even earlier than those we have already mentioned. These are the writings of “The Mechanical School of Alexandria” that was active in the third century BCE. These were written in Greek but most of them have survived only in Arabic and in the past this led many scholars to the conclusion that the references to water mills are later Arabic interpolations and therefore they did not take them into account. Recently, however M. J. T. Lewis (1997) has made a detailed study of these and related documents and has come to the conclusion that the mention of water mills is part of the original text. He shows that a Greek papyrus of the 2^nd c. BC known as Laterculi Alexandrini which enumerates small lists of notable places and famous people mentions a man by the name of Abdaraxos and states that “he completed the machines of Alexandria” (Lewis 1997, p. 60). Louis then translates and interprets a section of a difficult Arabic text, apparently of Apollonius of Perga (262-190 BC), as referring to the same man. According to Lewis the section reads: “the Byzantine mill or the newly invented vertical mill of Abdaraxos” (Lewis 1997, p. 52, 60). Lewis suggests that the Byzantine mill is the horizontal mill and that Abderaxos is a Semitic name probably Phoenician. If Lewis’reading is correct this is evidence both for the horizontal mill’s being earlier than the vertical mill and also for where and when the vertical Vitruvian mill was invented, in Alexandria in the third century BCE. Remarkably it also provides the name and origin of the inventor.

9Very few ancient water mills have been excavated in Israel. One from’Ein Gedi, excavated in 1998 had a horizontal water wheel but no Aruba penstock and was dated to the Byzantine period (the equivalent of the Late Roman in the West). As a result Gideon Hadas (2001-2002) the excavator suggests that the Aruba penstock developed later. Mills found in surveys suggest that this is not correct but only further excavations can decide the matter.

10Several water mills have been excavated at Nahal Tanninim, “the Crocodile River” near the sea north of Caesarea, the Roman capital of Judea. Nahal Tanninim was dammed off by two dams to create a large artificial lake (area 500-600 ha.) to supply Caesarea with water. There were water mills of several periods on and near both dams. Yoseph Porat has excavated a mill on the northern dam that was used in three different periods but has not yet been published (Porath forthcoming). In 1984 Thorkild Schiøler excavated one mill on the western dam that he dated to 345-380 (Schiøler 1989) on the basis of C14 of the plaster. Schiøler suggested that the mill was originally a turbine mill, with the horizontal water wheel at the bottom of the wide tube and that only later was it converted to the more usual type with the water wheel next to the Aruba penstock. Also in 1984 John Peter Oleson published a rock-cut twin mill that he identified to the south west of the same dam (Oleson 1984). Five more mills were later found by Uzi Ad and’Abd al-Salam Sa’id in the same vicinity (Ad 2005). In the years 2000-2003 Ad and al-Salam Said excavated five of the six and dated them to the Byzantine period. Surprisingly the mills had vertical water wheels which is probably to be explained as being a result of Roman influence from Europe. Of greater importance is the fact that, to the best of our knowledge, these are the first mills to be discovered with. Pompeian type millstones as opposed to the usual disk shaped millstones that are similar to those of the rotary hand mill and are found in all water-mills both ancient and pre-industrial (for Pompeian mills see Moritz 1958, p. 74-96 and Peacock 1989). The lower millstones at Nahal Tanninim differ from those of the usual Pompeian donkey mill in that they have vertical bores for the spindles that turn the upper stones. The upper stones are also different from those of the usual Pompeian mill in that there is a depression on the rim for the rod to which the spindle was attached. In some cases there are still traces of the square mortices typical of the Pompeian mill but in other cases it is no longer there. After the discovery at Nahal Tanninim, Nimrod Getzov brought to our attention similar millstones at two museums in the north of the country, one at Kibbutz Nir Dawid near a river in the Jezreel Valley and the other at Kibbutz Ma’ayan Barukh near the sources of the Jordan (the photographs of the latter are published here for the first time with the kind permission of Amnon Asaph the curator of the museum). Previously the significance of these stones had not been understood. The upper millstones from Ma’ayan Barukh are different from those from Nahal Tanninim in that instead of the depression on the rim there are two projections. The rod to which the spindle was attached was probably tied to these projections.

11In order to fully understand the historical significance of the use of Pompeian type millstones in water mills it is necessary to give a short survey of the history of grinding stones. Early grinding stones everywhere were simple saddle querns. These are the millstones both of the Bible and of other early literature such as Homer At sometime in the fifth century BCE two different types of mills developed one in the eastern Mediterranean and the other in the western Mediterranean. The Olynthus lever mill developed somewhere in the eastern Mediterranean (Frankel 2003a). In the western Mediterranean at about the same time the rotary hand mill developed. It is known in Spain already in the fifth century BCE (Alonso Martinez 1995, 1996) Also in the west about a century later the Pompeian Donkey Mill developed from the rotary hand mill. An early type was found at Morgantina (White 1963) in Sicily, at Byrsa in North Africa (Lancel 1982) and on the El Sec shipwreck where it was clearly dated to the fourth century BCE (Arribas 1987). The Olynthus lever mill was brought to the west by Greek settlers but did not penetrate regions such as Spain or Northern France where the rotary hand-mill was well established. The rotary hand mill took much longer to reach the east and was apparently brought there mainly by the Roman legions. For example at Masada, the Jewish stronghold captured by the Romans in 73 CE (Josephus Flavius, The Jewish War, 7, 252-259; 275-406) the Jewish fighters still used Olynthus mills but the Roman soldiers used rotary hand mills (Netzer 1991, p. 290-291, ills. 464-465) and had no use for the Olynthus mills that the zealots had used and converted them to paving stones (ibid. ill. 890) The picture in Greece is similar (Runnels 1990), and in the east the Olynthus mill was replaced by the rotary hand mill only very gradually and apparently remained as the main hand-mill till well into the Roman period if not later. Therefore the hand mill of the Rabbinic works written in the Roman Period (eg the Mishna and Tosephta) and of the New Testament (eg Math. 24, 41) was the Olynthus mill. The Pompeian donkey mill, however, is found In the eastern Mediterranean already in the Hellenistic period (Magen 1993, fig. 38, 1) and is found in large numbers in the Roman Period. As a result in the west the rotary hand mill precedes the Pompeian Donkey Mill while in the east the reverse is true and the Pompeian Donkey mill appears first. The explanation is probably that in the east there was already a good hand mill so that the rotary hand mill was only accepted slowly but there was no donkey mill so that the Pompeian donkey mill was accepted much more rapidly.

12To the best of my knowledge all the water mills found so far every where, ancient and modern, except those from Nahal Tanninim, have disc shaped grinding stones similar to those of rotary hand mills If the water mill had reached the Levant from somewhere else it would almost certainly have had disc shaped millstones. The fact that the millstones from Nahal Taninnim are of the Pompeian type is almost certainly a survival from a period when rotary hand mills had not yet reached the eastern Mediterranean. They therefore conform well with the possibility that the water-mill originated in the eastern Mediterranean at a period between the arrival of the Pompeian Donkey mill and that of the rotary hand mill. Scholars such as Bennet and Elton and Avitsur suggested that the water mill was invented in the eastern Mediterranean and if Lewis’interpretation of the early Greek sources is accepted, this is reflected in these documents also. The discovery of Pompeian type millstones in the water mills at Nahal Tanninim provide additional support for this tenet.